Add a test for tffm0()

4 files changed, 41 insertions, 23 deletions

diff --git a/R/tffm0.R b/R/tffm0.R
index bb5f4cf5..56283a5d 100644
--- a/R/tffm0.R
+++ b/R/tffm0.R
@@ -1,15 +1,18 @@
 #' Transform formation fractions as in the first published mkin version
 #'
-#' The transformed fractions can be restricted between 0 and 1 in model
-#' optimisations. Therefore this transformation was used originally in mkin. It
-#' was later replaced by the [ilr] transformation because the ilr transformed
+#' This transformation was used originally in mkin, in order to implement a
+#' constraint for the sum of formation fractions to be smaller than 1.  It was
+#' later replaced by the [ilr] transformation because the ilr transformed
 #' fractions can assumed to follow normal distribution. As the ilr
 #' transformation is not available in [RxODE] and can therefore not be used in
-#' the nlmixr modelling language, this transformation is currently used for
-#' translating mkin models with formation fractions to more than one target
-#' compartment for fitting with nlmixr in [nlmixr_model]. However,
-#' this implementation cannot be used there, as it is not accessible
-#' from RxODE.
+#' the nlmixr modelling language, the original transformation is currently used
+#' for translating mkin models with formation fractions to more than one target
+#' compartment for fitting with nlmixr in [nlmixr_model]. However, this
+#' implementation cannot be used there, as it is not accessible from RxODE.
+#' 
+#' If the transformed formation fractions are restricted to the interval
+#' between 0 and 1, the sum of backtransformed values is restricted
+#' to this interval.
 #'
 #' @param ff Vector of untransformed formation fractions. The sum
 #'   must be smaller or equal to one
diff --git a/man/tffm0.Rd b/man/tffm0.Rd
index 46978d5e..89bee7d9 100644
--- a/man/tffm0.Rd
+++ b/man/tffm0.Rd
@@ -22,16 +22,20 @@ A vector of the transformed formation fractions
 A vector of backtransformed formation fractions for natural use in degradation models
 }
 \description{
-The transformed fractions can be restricted between 0 and 1 in model
-optimisations. Therefore this transformation was used originally in mkin. It
-was later replaced by the \link{ilr} transformation because the ilr transformed
+This transformation was used originally in mkin, in order to implement a
+constraint for the sum of formation fractions to be smaller than 1.  It was
+later replaced by the \link{ilr} transformation because the ilr transformed
 fractions can assumed to follow normal distribution. As the ilr
 transformation is not available in \link{RxODE} and can therefore not be used in
-the nlmixr modelling language, this transformation is currently used for
-translating mkin models with formation fractions to more than one target
-compartment for fitting with nlmixr in \link{nlmixr_model}. However,
-this implementation cannot be used there, as it is not accessible
-from RxODE.
+the nlmixr modelling language, the original transformation is currently used
+for translating mkin models with formation fractions to more than one target
+compartment for fitting with nlmixr in \link{nlmixr_model}. However, this
+implementation cannot be used there, as it is not accessible from RxODE.
+}
+\details{
+If the transformed formation fractions are restricted to the interval
+between 0 and 1, the sum of backtransformed values is restricted
+to this interval.
 }
 \examples{
 ff_example <- c(
diff --git a/test.log b/test.log
index a2217f49..cb7edba3 100644
--- a/test.log
+++ b/test.log
@@ -8,19 +8,19 @@ Loading required package: parallel
 ✔ |         2 | Export dataset for reading into CAKE
 ✔ |        12 | Confidence intervals and p-values [1.0s]
 ⠋ |         1 | Dimethenamid data from 2018                                     
-✔ |     1  25 | Dimethenamid data from 2018 [49.4s]
+✔ |     1  25 | Dimethenamid data from 2018 [52.3s]
 ────────────────────────────────────────────────────────────────────────────────
 Skip (test_dmta.R:147:3): Different backends get consistent results for SFO-SFO3+, dimethenamid data
 Reason: Fitting this ODE model with saemix takes about 15 minutes on my system
 ────────────────────────────────────────────────────────────────────────────────
-✔ |        14 | Error model fitting [6.9s]
+✔ |        14 | Error model fitting [7.1s]
 ✔ |         5 | Time step normalisation
 ✔ |         4 | Calculation of FOCUS chi2 error levels [0.8s]
 ✔ |        14 | Results for FOCUS D established in expertise for UBA (Ranke 2014) [1.2s]
 ✔ |         4 | Test fitting the decline of metabolites from their maximum [0.6s]
 ✔ |         1 | Fitting the logistic model [0.4s]
 ⠴ |         6 | Nonlinear mixed-effects models                                  
-✔ |     1  15 | Nonlinear mixed-effects models [2.4s]
+✔ |     1  15 | Nonlinear mixed-effects models [2.5s]
 ────────────────────────────────────────────────────────────────────────────────
 Skip (test_mixed.R:67:3): saemix results are reproducible for biphasic fits
 Reason: Fitting with saemix takes around 10 minutes when using deSolve
@@ -30,24 +30,25 @@ Reason: Fitting with saemix takes around 10 minutes when using deSolve
 ✔ |         3 | mkinfit features [1.1s]
 ✔ |         8 | mkinmod model generation and printing [0.2s]
 ✔ |         3 | Model predictions with mkinpredict [0.3s]
-✔ |        16 | Evaluations according to 2015 NAFTA guidance [2.1s]
+✔ |        16 | Evaluations according to 2015 NAFTA guidance [2.2s]
 ✔ |         9 | Nonlinear mixed-effects models with nlme [9.2s]
 ✔ |        16 | Plotting [1.5s]
 ✔ |         4 | Residuals extracted from mkinfit models
-✔ |        23 | saemix_parent [29.2s]
+✔ |        23 | saemix_parent [29.1s]
 ✔ |         2 | Complex test case from Schaefer et al. (2007) Piacenza paper [1.6s]
 ✔ |         7 | Fitting the SFORB model [4.4s]
 ✔ |         1 | Summaries of old mkinfit objects
 ✔ |         4 | Summary [0.1s]
 ✔ |         4 | Results for synthetic data established in expertise for UBA (Ranke 2014) [2.6s]
 ✔ |         9 | Hypothesis tests [9.4s]
+✔ |         2 | tffm0
 ✔ |         4 | Calculation of maximum time weighted average concentrations (TWAs) [2.2s]
 
 ══ Results ═════════════════════════════════════════════════════════════════════
-Duration: 132.0 s
+Duration: 135.1 s
 
 ── Skipped tests  ──────────────────────────────────────────────────────────────
 • Fitting this ODE model with saemix takes about 15 minutes on my system (1)
 • Fitting with saemix takes around 10 minutes when using deSolve (1)
 
-[ FAIL 0 | WARN 0 | SKIP 2 | PASS 237 ]
+[ FAIL 0 | WARN 0 | SKIP 2 | PASS 239 ]
diff --git a/tests/testthat/test_tffm0.R b/tests/testthat/test_tffm0.R
new file mode 100644
index 00000000..82e1ee94
--- /dev/null
+++ b/tests/testthat/test_tffm0.R
@@ -0,0 +1,10 @@
+test_that("The formation fraction transformation tffm0 is reversible", {
+  ff_example <- c(
+    0.10983681, 0.09035905, 0.08399383
+  )
+  ff_example_trans <- tffm0(ff_example)
+  expect_equal(invtffm0(ff_example_trans), ff_example)
+
+  ff_ex_2_trans <- c(0.5, 0.9, 0.99)
+  expect_true(sum(invtffm0(ff_ex_2_trans)) < 1)
+})